"Dark GPUs" at Amazon show the limits of the AI buildout: chips are easy to buy, power is not
AWS is reportedly buying billions in Nvidia GB300 racks every month and storing them in warehouses
Jensen Huang says Nvidia's AI chips carry "the lowest" risk of any asset and could serve as collateral for debt
Huang says AI chips stay useful for at least eight years, Elon Musk is making a similar case
Nvidia reportedly refused some neoclouds access to GTC Berlin for adding AMD and TPU chips
SemiAnalysis says Nvidia pressure is real, but the evidence so far is unnamed neoclouds
Nvidia GeForce drivers add support for Microsoft's Advanced Shader Delivery
Nvidia's latest Game Ready driver adds support for Microsoft's Advanced Shader Delivery, which aims to reduce shader compilation stuttering by delivering precompiled shaders for supported games, including Call of Duty: Modern Warfare 4 and Gears of War: E-Day. As our recent testing shows, shader stuttering remains a major headache for PC gamers, but efforts to eliminate it are gaining momentum.
DLSS 5 Just Got a Lot Faster, but Is It Worth Using?
Nvidia's new DLSS 5 model nearly doubles neural rendering speed across the RTX 50 series. We put it to the test to see if the performance gains outweigh the heavy FPS penalty and lingering visual issues.
Nvidia's RTX 6060 probably won't arrive until 2029, will be slower than the PS6
Memory shortages will keep delaying mainstream GPUs
Rumor mill: Rumors had already suggested that the AI business and related memory shortages would significantly widen the traditional gap between generations of Nvidia graphics cards. Now, the latest information suggests the situation could be even worse for upcoming mainstream GPUs, which probably won't match the performance of next-generation consoles.
Can a Simple App Really Fix PC Game Stutters?
Shader compilation stutter has plagued PC gaming for years, causing frustrating hitches even on powerful hardware. The SCSKiller app promises to do something about it, but does it work?
This mod lets Nvidia GeForce owners run a custom VBIOS without flashing the card
BIOS modding for GeForce cards is back, until Nvidia patches it
Reseller says $470,000 worth of Nvidia RTX 5090s went missing at São Paulo airport
The airport, the Federal Police, and the Federal Revenue Service haven't commented, and no one has officially confirmed a theft
Nvidia says the RTX 5090 isn't discontinued, but most listings are over $9,300
New rumors point to a RTX 5080 24GB Super replacement
First Surface Laptop Ultra reviews: a promising but pricey MacBook Pro rival
Microsoft Surface Laptop Ultra reviews have landed with a mixed verdict. Powered by RTX Spark, the premium Arm laptop earns praise for its MacBook-like build quality, AI capabilities, and versatility. However, its steep price, modest battery life, and Windows software limitations keep it from fully delivering on its ambitious promises.
Nvidia and Cisco unveil next-gen network interface built for the AI data center boom
Four times the bandwidth, much better power efficiency for new "AI factories"
Nvidia Broadcast 2.3 adds 4K video upscaling and frame generation
Nvidia Broadcast has received a major upgrade, bringing AI-powered 4K video upscaling and frame generation to webcams. The update also moves beyond live streaming, allowing users to enhance previously recorded videos and audio files. The features require compatible RTX GPUs.
Is the Radeon RX 9070 XT Really Faster Than the RTX 5090?
Can a Radeon RX 9070 XT really beat Nvidia's flagship RTX 5090? We tested 11 games at 1440p and 4K to see when driver overhead matters, and when an ordinary CPU bottleneck is actually to blame.
Nvidia says DLSS 5 is now twice as fast and uses 20% less VRAM
DLSS frame generation also receives a lighter, faster preset
Nvidia may be killing off the RTX 5090 as AI chips take priority over gaming GPUs
"Greed, for lack of a better word... is good."
From gaming graphics to AI dominance: Nvidia's Jensen Huang receives America's highest science honor
The award recognizes his role in advancing massively parallel computation
Nvidia and Microsoft say the "agentic PC" era starts with the Surface Laptop Ultra
Windows gets Actions in Search and a hybrid AI toggle as Nvidia's RTX Spark laptops arrive
Microsoft's Surface Laptop Ultra starts at $2,600, running Nvidia's RTX Spark chip
The 15-inch 4K touchscreen goes up to 120Hz, with up to 128GB of unified memory and a magnetic 140W USB-C charging port
Something to look forward to: Microsoft has just revealed the starting price and release date for its new ultra-tier Surface laptop, along with other notebooks powered by Nvidia's RTX Spark SoC. The company's flagship model is expensive, even next to the MacBook Pro it's meant to rival, but it remains to be seen how RTX Spark stacks up against Apple's M5 Pro.
Microsoft's Windows and Surface event starts at 1 pm ET, watch it live for more on RTX Spark
Expect details on the Surface Laptop Ultra, RTX Spark hardware, Windows AI features, and release dates
Gears of War: E-Day GPU Benchmark: 44 GPUs Tested
We tested 44 GPUs in Gears of War: E-Day at 1080p, 1440p, and 4K to see what it takes to run the game well, how 8GB cards hold up, and whether upscaling makes Ultra settings practical.
RTX 5090 modder claims to have solved melting power connectors with auxiliary 8-pin board
The mod spreads power draw across three sets of cables
Microsoft's fix for shader compilation wait is coming to Nvidia, Intel, and Qualcomm GPUs
Advanced Shader Delivery sends precompiled shaders from the cloud
Nvidia rolls out security update for GeForce GTX 700, 900 and 10 series GPUs
Nvidia has released GeForce 582.78 WHQL, a security update for older Maxwell, Pascal, and Volta GPUs that no longer receive regular Game Ready drivers. The release includes security fixes but no new game optimizations or features. Nvidia plans to provide critical security updates for these architectures through October 2028.
Ex-Nvidia advisor says the company owes him $1 billion in stock, but Nvidia disagrees
A reported 1996 paperwork error miscounted his vesting period as four years instead of four quarters